The Number Getting Quietly Buried

Everyone knows the job displacement narrative. AI takes tasks, tasks become jobs, jobs disappear. The story has been told so many times that a counterintuitive data point barely registers when it surfaces.

The World Economic Forum: AI has already created more than 1.3 million new jobs globally , including roles that either did not exist or existed in negligible numbers before 2023.

That is not a prediction. It is a measurement of what has already happened.

The Specific Numbers

LinkedIn's 2026 Jobs on the Rise report ranked AI Engineer as the number one fastest-growing job title in the United States. Year-over-year growth: 143%.

Four of LinkedIn's top five fastest-growing job titles are AI-related. AI and machine learning job postings surged 163% from 2024 to 2025.

Three years ago, prompt engineer was not a job title. It now has a 135.8% growth rate and a defined career path. AI trainers , people who evaluate and correct AI-generated content to improve model performance , are a growing category. AI governance specialists. Model evaluators. Roles that require human judgment, domain expertise, and contextual understanding that the technology itself does not have.

We invented a technology so powerful that it requires an entirely new profession dedicated entirely to asking it questions in the right way. Whether that is a breakthrough or the most expensive autocomplete in history is an open question. But the jobs are real.

Why This Keeps Surprising Economists

In 1865, William Stanley Jevons observed something strange. Steam engines had become dramatically more efficient , you needed less coal to produce the same energy. Total coal consumption went up. Massively. Because cheaper, more efficient energy unlocked new uses. New industries became viable. The efficiency gains did not reduce demand for coal. They expanded it.

Apollo's chief economist has argued the same logic applies to AI. If AI makes certain types of work cheaper and faster, the demand for that work does not disappear , it expands. New applications become viable. New roles emerge around managing, correcting, and directing the technology.

This is the Jevons Paradox. The Victorians discovered it by accident with coal. AI is running the same experiment right now, and the early data suggests the same dynamic is playing out.

The Jobs That Are Actually Growing

The growth is not random. It clusters around a specific type of work: human judgment applied to AI outputs.

Prompt engineers design the inputs that produce reliable outputs from AI systems. AI trainers evaluate outputs and improve model performance through feedback. Governance specialists ensure AI systems do not produce catastrophic results. Model evaluators assess whether outputs meet the standards required for deployment.

None of these roles require you to build AI from scratch. All of them require understanding how AI systems fail , and why. Domain expertise matters. Contextual judgment matters. The ability to tell the difference between an output that looks correct and one that actually is correct matters.

These are not temporary bridge jobs. They exist because the technology genuinely requires them. The companies that recognised this early are the ones showing high ROI from AI. The companies that tried to eliminate the human layer entirely are the ones in the Gartner study that cut 80% of their pilot teams and saw no improvement in returns.

The job market is changing. It is not contracting in the way the displacement narrative describes. It is restructuring around a different kind of value , and the people who understand that restructuring early have a real edge on the people still waiting for the ceiling to fall.